bhearn26/Nemotron-Personas-USA

Nemotron-Personas-USA是一个开源（CC BY 4.0）数据集，包含基于真实世界人口统计、地理和人格特质分布生成的合成人物角色，以捕捉美国人口的多样性和丰富性。它是首个与姓名、性别、年龄、背景、婚姻状况、教育、职业和位置等属性统计信息对齐的数据集。数据集可用于提高合成生成数据的多样性、减轻数据/模型偏见并防止模型崩溃。它通过NVIDIA的NeMo Data Designer生成，结合了概率图模型和开源的openai/gpt-oss-120b模型。数据集包含100万条记录，每条记录有6个人物角色字段（如专业、体育、艺术、旅行、烹饪和综合人物角色）和16个上下文字段（如文化背景、技能、爱好、职业目标、年龄、婚姻状况、教育水平、职业、地理位置等），总计约9.36亿个词元。数据基于美国人口普查和美国社区调查等公开统计数据进行种子构建，覆盖美国50个州及波多黎各、维尔京群岛，包含56万多个职业类别和97万个唯一全名。数据集仅包含成年人，旨在为训练大型语言模型、主权AI开发等用例提供高质量、多样化的人物角色数据。

Nemotron-Personas-USA is an open-source (CC BY 4.0) dataset of synthetically-generated personas grounded in real-world demographic, geographic and personality trait distributions to capture the diversity and richness of the population. It is the first dataset of its kind aligned with statistics for names, sex, age, background, marital status, education, occupation and location, among other attributes. With an initial release focused on the United States, this dataset provides high-quality personas for a variety of modeling use-cases. The dataset can be used to improve diversity of synthetically-generated data, mitigate data/model biases, and prevent model collapse. Produced using NeMo Data Designer, an enterprise-grade compound AI system for synthetic data generation, the dataset leverages a proprietary Probabilistic Graphical Model (PGM) along with an Apache-2.0-licensed openai/gpt-oss-120b model. The dataset contains 1M records with 6 persona fields and 16 contextual fields, totaling ~0.94B tokens. Seed data is based on open-source aggregated statistical data from the US Census Bureau and other sources, covering 50 states plus Puerto Rico and Virgin Islands, with 560+ professional occupations and 970k unique full names. The dataset is focused on adults only and is intended for use in training LLMs, Sovereign AI, and improving synthetic data diversity.